Correction - PNAS · Chlorophyll and other pigments absorb light, which excites electrons. These electrons are passed along a chain of molecules, which use them to pry water molecules

NEWS FEATURE

Liquid sunlightFuels created by artificial photosynthesis are getting much closer to reality.

Katherine Bourzac, Science Writer

On a bench in a laboratory at Berkeley sits a devicedesigned to make the ultimate green fuel. The stain-less steel chamber, festooned with gaskets, nuts andbolts, and glass windows, looks like some kind ofsteam-punk aquarium. Inside, arrays of nanowire elec-trodes and bacterial colonies are using the light to turnwater and carbon dioxide into methane, the maincomponent in natural gas. This is one of the best at-tempts yet to realize the simple equation: sun+water+carbon dioxide = sustainable fuel.

Solar power is already a success story, but theelectricity generated by photovoltaic panels can’t doevery job. It is not much use in the tank of an airplane,for example. Three-quarters of the energy people usetoday is in the form of liquid and gaseous fuels, so therenewable energy portfolio needs fuels too.

That is why chemists are trying to copy plants.Through photosynthesis, plants take in carbon dioxide,water, and sunlight, and turn it into the chemicalsthey need, with oxygen as the only byproduct. Fordecades, scientists have wondered: Can we take aleaf out of the plant’s playbook, and grow our fuelsand chemicals?

Last year brought cause for optimism, as re-searchers made three advances toward practical solarfuels. “When people see what we’ve done, they willrealize it’s not pie-in-the-sky anymore,” says DanielNocera, a professor of chemistry at Harvard University.Now with the implications of climate change looming,competing researchers are advancing their prototypesand racing to not only prove technological success butalso show marketplace viability.

Planting the SeedsThe first step in the complex chemistry performed byplants is to split water into hydrogen and oxygen.Chlorophyll and other pigments absorb light, whichexcites electrons. These electrons are passed alonga chain of molecules, which use them to pry watermolecules apart.

Splitting water is also central to artificial photo-synthesis. This can be an end in itself, because hy-drogen can be used as a fuel or it can be a first steptoward more energy-dense hydrocarbon fuels, such asmethane and ethanol.

Researchers have been working to make solar fuelssince the 1970s. The inspiration came in 1972, whenAkira Fujishima at Kanagawa University and Kenichi

Honda at the University of Tokyo showed that twoelectrodes—one titanium dioxide and the other plat-inum—would catalyze the splitting of water when il-luminated with visible light (1).

In Fujishima and Honda’s system, photons hit thetitanium dioxide and create pairs of negative andpositive charges: electrons and holes. The electronsflow through a wire to the platinum electrode, whereasthe holes grab fresh electrons from water moleculesat the surface of the titanium electrode, splitting the

Researchers are racing to achieve artificial photosynthesis, hoping to generate asustainable liquid fuel directly from the sun. Image courtesy of Dave Cutler.

www.pnas.org/cgi/doi/10.1073/pnas.1604811113 PNAS | April 26, 2016 | vol. 113 | no. 17 | 4545–4548

NEW

SFEATURE

Dow

nloa

ded

by g

uest

on

June

24,

202

0 D

ownl

oade

d by

gue

st o

n Ju

ne 2

4, 2

020

Dow

nloa

ded

by g

uest

on

June

24,

202

0

molecules into hydrogen ions and oxygen. The hydro-gen ionsmigrate through the liquid to the platinum site,where they complete the circuit and recombine withelectrons to form molecules of H2.

Along with the oil crisis of 1973, this work inspiredmany young scientists to work on artificial photosyn-thesis. Arthur Nozik was among them. “I saw thatpaper and I got interested in solar power,” he recalls.Nozik was one of the founding researchers at whatwould become the National Renewable Energy Lab-oratory in Boulder, Colorado, where he began work-ing on new electrode designs for water splitting.

Hydrogen HopesThis first wave of enthusiasm soon passed as the priceof oil came down and the budget for renewable en-ergy research was cut during the Reagan administra-tion. But Nozik and a few others kept the flame alight.

Then in 1998, John Turner at the National Renew-able Energy Laboratory provided a sign that this workwas paying off, with an electrode system that couldsplit water with 12.4% efficiency (2). This was anotherturning point, and as the risks of climate change

became clearer in the early 2000s, more researchersjumped back in.

One of the first aims was to find an alternative toexpensive platinum electrodes. So researchers havebeen working to squeeze higher efficiency out ofmore abundant materials, including nickel and mo-lybdenum sulfides. The Joint Center for ArtificialPhotosynthesis (JCAP), a Department of Energyprogram housed at the California Institute of Tech-nology (Caltech), has tested hundreds of thousandsof new catalysts, and their results are promising.Some of their discoveries match the performance ofplatinum; one of the best is a compound of cobaltand molybdenum (3).

These catalysts are now being used in a slightlydifferent approach from Fujishima and Honda’s. In-stead of using light directly, water can be split byplugging electrodes into a source of electrical power.The current then drives the same reactions that wereset off by the charge-splitting effect of the photons.And if you generate that electrical power using a solarcell, you have a renewable source of fuel.

In August (4), chemist Leone Spiccia at MonashUniversity in Victoria, Australia, demonstrated such atwo-part system that could possibly compete in atough market; it also broke efficiency records. Spicciaused high-performance triple-junction solar cells togenerate electricity. The electricity passes throughnickel-foam electrodes to catalyze water splitting. Thesystem converts solar energy into hydrogen fuel withan efficiency of 22%. Spiccia is now working on re-ducing inefficiencies in the connections between theparts, and he believes that an overall efficiency of 28%or 30% is possible.

But being green is only one argument for a tech-nology; it also has to make economic sense. Triple-junction solar cells are expensive, so Spiccia’s systemmight need subsidies to compete with dirtier sourcesof hydrogen.

Today, hydrogen is primarily made by steamreforming of methane, an energy-intensive but in-expensive process. Hydrogen made this way costsabout $2 per kilogram, says Nathan Lewis, a chemistat Caltech and the former director of JCAP at Caltech.Making hydrogen using electrolysis fed by conven-tional solar cells would come in at around $5 to $7 perkilo, he estimates. Spiccia hasn’t done a full cost anal-ysis but readily admits hydrogen made using his pro-totype would be more expensive than what is on themarket today. There’s much room for improvement inhis first demo system.

Lewis favors a design that eliminates the need for aseparate solar cell. As part of JCAP, he developed awater-splitting system with electrodes that are some-thing like submerged photovoltaic panels. His systemlooks like a sealed reactor full of water, illuminatedfrom the outside, shiny photodiodes within. As in anordinary solar cell, light strikes a semiconductor, gen-erating electrons and positively charged “holes.” Butrather than funnel these off to an electrical grid or abattery, the JCAP device passes them directly to cat-alysts to split water.

This schematic shows the basic approach of artificial photosynthesis projects be-ing pursued by the US Department of Energy-funded JCAP. A top membraneabsorbs light, CO2, and water while allowing oxygen to escape. Selectedmolecules embedded in an inner membrane catalyze reactions to produce fuel.The base layer wicks the fuel away. Image courtesy of the Joint Center for ArtificialPhotosynthesis, copyright Caltech.

4546 | www.pnas.org/cgi/doi/10.1073/pnas.1604811113 Bourzac

Dow

nloa

ded

by g

uest

on

June

24,

202

0

One of Lewis’s main challenges has been making asolar cell that will not break down underwater. The keyto this was a thin protective layer of titanium dioxide afew nanometers thick. In August, the group publishedthe results of their reactor design, which can producehydrogen fuel with 10% efficiency (5).

Lewis explains his long-term vision for hydrogenproduction: a system that would use printable ma-terials to make large-area, flexible reactors that canbe deployed cheaply. “We want to make somethingsimple enough to spray onto your house,” he says.That ultimate goal is still a big basic materials scienceand research problem.

In the meantime, Lewis is motivated by tryingto get something realistic to market as soon as pos-sible, he says. For him, that’s a solar fuel system thatmakes hydrogen.

Hydrogen isn’t the ideal fuel, as almost all ourexisting infrastructure is built for more energy-denseoptions, like gasoline and methane. One immediatebenefit of having a clean source of hydrogen would befor sustainable production of ammonia for fertilizer,which is made by combining nitrogen and hydrogen.Hydrogen can also power fuel cells, and above all, it canbe used as a starting point for other reactions. “I think ofhydrogen as a way to upgrade things,” says Jens Nor-skov, professor of photon science at Stanford University.

Green GasStill, it would be more efficient if an artificial leaf couldproduce more energy-dense fuels directly, by usingcarbon dioxide as a feedstock. Carbon dioxide can becaptured from power plants, and the aim of manyprojects is to then store the gas. It would be muchmore useful to convert the stuff into a transportationfuel or a high-value chemical.

Harry Atwater, now director of JCAP, says metha-nol or ethanol would be good options. Ethanol is al-ready blended into fuel, and there are efficient ways toconvert methanol into gasoline. But generating eventhese relatively simple hydrocarbons is much harderthan splitting water.

That’s because the chemistry is much more com-plex. Splitting a molecule of water takes four electrons,says Norskov. Making the simple hydrocarbon methaneis a reaction involving eight electrons, each with differ-ent energies, which have to be shuffled around throughseveral steps to create the single-carbon molecule.

What chemists can’t easily do in the laboratory,leaves do with ease: making complex sugars and otherorganic molecules. Nature uses 3D enzymes to wran-gle all of the ingredients, roping them together tomake all of the intermediate reactions and electrontransfers happen in order. These delicate natural cat-alysts are rapidly damaged by the energetic process,and are nearly continuously rebuilt and replaced byplant cells. Synthetic catalysts must either heal them-selves somehow—an idea Nocera has been workingon—or be incredibly durable, made out of hard ma-terials. Designing a self-healing or durable catalystthat can pull off all this chemistry is tremendously

challenging. “There’s nothing that works even close towell enough,” says Norskov.

Perhaps the greatest challenge for constructing theartificial leaf is matching the specificity of plant en-zymes. The natural proteins can produce very specificproducts, such as pure methane, whereas syntheticcatalysts tend to churn out an unpredictable medley ofcarbon-containing compounds.

Building with BacteriaInstead of waiting for synthetic chemists to match na-ture’s wonders, some chemists have recruited bacteriato help do the job. Peidong Yang, at the University ofCalifornia, Berkeley, has made a complete solar fuelssystem with what he calls living catalysts. His systemuses nanowire electrodes, which are in principle similarto Fujishima andHonda’s work, and Turner’s as well. Butbecause they have nanowire-carpeted surfaces ratherthan smooth ones, these electrodes can both absorbmore light and hold more catalyst in a given area thanearlier ones. That means they can better keep up withsolar flux, and get the energy from more photons con-verted into more electrons that can split more water.One electrode has nanowires paired with a syntheticcatalyst; another is seeded with bacteria.

A design published recently in PNAS makes meth-ane (6). This set-up uses a double-chambered reactor.

“I don’t know whether a synthetic catalyst or a bacterialone will win out.”

—Peidong Yang

In one windowed chamber, the anode splits waterto make oxygen gas and hydrogen ions, the samebasic process other researchers have exploited. In asecond chamber, a cathode coated with a nickel cat-alyst brokers a reaction between the hydrogen ionsand electrons to make dissolved hydrogen gas.Methanosarcina barkeri bacteria, acting as living cat-alysts, take up that gas and combine it with CO2 tomake methane. The process is highly efficient: 86% ofthe electrons produced by splitting water are used inthe methane-producing reaction. The methane bub-bles out of the water and is then captured. It’s tooearly to make firm predictions about the commercialcost of a large-scale system of this kind, but bacteriaare already routinely used to brew alcohol and evendrugs in large vats. And there is a product that uses ananalogous combination of human chemistry andbacterial smarts, the semisynthetic malaria drug arte-misinin made by Sanofi (7).

Another of Yang’s designs uses a strain of bacteriathat produces acetate. That acetate is then eaten bygenetically engineered Escherichia coli that can con-vert it into plastics or butanol. Yang is collaboratingwith synthetic biologist Michelle C. Y. Chang, who isdeveloping strains of bacteria that can both generatea greater variety of chemicals and live in the reactor.At Harvard, Nocera and synthetic biologist Pamela

Bourzac PNAS | April 26, 2016 | vol. 113 | no. 17 | 4547

Dow

nloa

ded

by g

uest

on

June

24,

202

0

Silver are also working on a design that uses microbes(8). Their efficiency is better; so far they’ve made iso-propanol but they’re also working on more widelyused fuels.

The Price of Sustainable FuelNorskov and others are excited to see these new ideasachieving success in prototypes. Just as researchershave had to move on from platinum catalysts, theymay end up using microbial partners for complete ar-tificial photosynthesis, if that turns out to work better.

“I don’t know whether a synthetic catalyst or abacterial one will win out,” says Yang. A chemist atheart, he favors an all-inorganic system. Bacteria aresensitive to pH, temperature, and other environmen-tal factors, all of which puts a certain strain on thedesign of the engineered components. But bacte-ria can do chemistry that synthetic catalysts can’t, soit’s worth babying them. Today bacteria are the best,

he says. Maybe they’ll be a stepping stone to a fullysynthetic system.

These three recent successes—Spiccia’s record-breaking electrolysis, JCAP’s integrated cell, andYang and Nocera’s full photosynthesis systems—arecause for hope. Still, demonstrating a laboratory pro-totype is very different from confronting the complexeconomic realities of the energy markets. If and whensolar fuels are first introduced, they are sure to bemore expensive than fossil fuels, so researchers mayneed to show that costs can come down further beforecompanies get involved.

Beyond the basic science, it will be a question ofpolitical will, as climate change policy remains con-tentious. “We’re going to need some help on thepolicy side,” says Atwater. A carbon tax would help, aswould subsidies for companies interested in commer-cializing these technologies. “The science and the pol-itics are mixed together,” says Nozik. “The question is:are we going to run out of time?”

1 Fujishima A, Honda K (1972) Electrochemical photolysis of water at a semiconductor electrode. Nature 238(5358):37–38.2 Khaselev O, Turner JA (1998) A monolithic photovoltaic-photoelectrochemical device for hydrogen production via water splitting.Science 280(5362):425–427.

3 McCrory CCL, et al. (2015) Benchmarking hydrogen evolving reaction and oxygen evolving reaction electrocatalysts for solar watersplitting devices. J Am Chem Soc 137(13):4347–4357.

4 Bonke SA, et al. (2015) Renewable fuels from concentrated solar power: Towards practical artificial photosynthesis. Energy Environ Sci8(9):2791–2796.

5 Verlage E, et al. (2015) A monolithically integrated, intrinsically safe, 10% efficient, solar-driven water-splitting system based on active,stable earth-abundant electrocatalysts in conjunction with tandem III–V light absorbers protected by amorphous TiO2 films. EnergyEnviron Sci 8(11):3166–3172.

6 Nichols EM, et al. (2015) Hybrid bioinorganic approach to solar-to-chemical conversion. Proc Natl Acad Sci USA 112(37):11461–11466.7 Bourzac K, Ross V (2014) Biologists modify yeast to produce malaria drug. Available at discovermagazine.com/2014/jan-feb/23-synthesizing-supply-for-malaria-drug. Accessed March 28, 2016.

8 Tortella JP, et al. (2015) Efficient solar-to-fuels production from a hybrid microbial-water-splitting catalyst system. Proc Natl AcadSci USA 112(8):2337–2342.

4548 | www.pnas.org/cgi/doi/10.1073/pnas.1604811113 Bourzac

Dow

nloa

ded

by g

uest

on

June

24,

202

0

Correction

NEWS FEATURECorrection for “News Feature: Liquid sunlight,” by KatherineBourzac, which appeared in issue 17, April 26, 2016, of Proc NatlAcad Sci USA (113:4545–4548; 10.1073/pnas.1604811113).The editors note that on page 4547, right column, last paragraph,

line 1, “Another of Cui’s designs” mistakenly appeared. The textshould have read “Another of Yang’s designs.” The online versionhas been corrected. We regret the error.

www.pnas.org/cgi/doi/10.1073/pnas.1606805113

E3048 | PNAS | May 24, 2016 | vol. 113 | no. 21 www.pnas.org